1. The value assumption

Philosophical discussions of Platonic value problems invariably treat them as problems about the comparative value of states. However, philosophers do not always formulate Platonic value problems in this way. They are often introduced as problems about our desires, as problems about our preferences, and as problems about our aims. As well as asking why knowledge is more valuable than true belief (a question about the comparative value of the state of knowledge), philosophers also ask why we desire knowledge and not just true belief, why we prefer knowledge to true belief, and why enquiry aims at knowledge and not just true belief.Footnote ⁴ As we will see in more detail in a moment, these different ways of formulating Platonic value problems are not equivalent. First, however, I want to substantiate the assertion that philosophers formulate value problems in these different ways, by giving some illustrative examples.

The first example is from Alan Millar's paper ‘Why Knowledge Matters’. In this paper Millar provides the following account of the question that interests him.

Millar subsequently assimilates his question to the Meno problem.Footnote ⁶ It is striking, though, that the issue Millar describes in the second sentence of the quoted passage isn't an issue about the comparative value of states, but rather one about the aim of an activity. What ‘philosophical tradition urges us to consider’, he says, is why knowledge and not ‘some state falling short of knowledge but implicating true belief’ is the goal of enquiry.Footnote ⁷ Although he immediately reformulates his problem as one about value (‘what is so good about grasping the truth in the way one does when one knows?’), his initial formulation is not in these terms.

Two other aspects of Millar's question about enquiry deserve comment. The first is the question's presupposition that knowledge is the aim, or goal, of enquiry. Even if one thinks there is something right about this idea, articulating what is right about it is not entirely straightforward. It certainly will not do to say, as Millar suggests, that ‘if the inquiry is genuinely into whether such-and-such is so then the aim is find [sic] out and thus come to know the right answer’ (64, original emphasis). One reason for this is that it is possible to enquire into a question using certain methods even if one knows that one cannot find out the answer to that question by those methods. A scientist can enquire into the colour of a dinosaur species by considering, e.g., the coloration that would have been most advantageous for the species given what we know of its lifestyle, despite the fact that scientists are well aware that they cannot conclusively establish the colours of dinosaurs using such methods.Footnote ⁸ It seems to me that the scientist engaged in this investigation is ‘genuinely’ enquiring into what colour the dinosaur species was, even if her aim cannot be to find out what colour the species was. A more plausible conception of her aim is that it is to determine what colour the species was most likely to have been, given (some of) the available evidence.Footnote ⁹

The second feature of Millar's question worth commenting on is that it asks not why enquiry aims at knowledge rather than true belief, but why enquiry aims at knowledge ‘rather than some state falling short of knowledge but implicating true belief’. This broadening of the question to comprise states besides true belief simpliciter represents a post-Platonic innovation. One effect of the obsession with analysing knowledge that gripped epistemology in the latter half of the twentieth century was to highlight the existence of a range of states allegedly intermediate between true belief and knowledge, the obvious example being that of having a justified true belief.Footnote ¹⁰ It is now widely assumed that an adequate account of the value of knowledge should explain not only why knowledge is more valuable than true belief, as Plato asked, but also why knowledge is more valuable than these intermediate states. The latter problem has become known as the ‘secondary value problem’, in contrast to the ‘primary value problem’ (i.e. the Meno problem) raised by Plato. Millar's question about enquiry is meant to encompass both issues.

Although there is a distinction to be drawn between the Meno problem and the secondary value problem, the latter remains a recognisably Platonic problem in that it concerns the greater value of knowledge in comparison to true belief-involving states insufficient for knowledge. All it adds to Plato's original conception of the problem is the observation that the range of true belief-involving states insufficient for knowledge extends beyond true belief simpliciter. This sharpens the problem of explaining the value of knowledge, because an account of the value of knowledge might succeed in explaining why knowledge is more valuable than true belief simpliciter without explaining why knowledge is more valuable than states intermediate between true belief and knowledge. An account of the value of knowledge might solve the primary value problem without solving the secondary value problem. However, the fundamental character of the two problems is the same. For this reason, I refer to any value problem that asks us to explain why knowledge is more valuable than true belief-involving states insufficient for knowledge as a Platonic value problem.Footnote ¹¹

The next two examples of how philosophers formulate Platonic value problems can be dealt with more quickly. In their entry on the value of knowledge for the Stanford Encyclopedia of Philosophy, Pritchard and Turri offer the following account of the secondary value problem:

Pritchard and Turri initially present the secondary value problem as an issue about the comparative value of the state of knowledge: the problem is to explain ‘why knowledge is more valuable than any proper subset of its parts’. However, they then reformulate the problem as a problem about our desires: why do we want to know, and not just to be in true belief-involving states falling short of knowledge? They appear to see no distinction between these two questions – or, at least, no distinction worth commenting on.

My final example is taken from Patrick Bondy's entry on ‘Epistemic Value’ for the Internet Encyclopedia of Philosophy.Footnote ¹³ Bondy offers the following account of the Meno problem:

Although Bondy refers to the Meno problem, following Pritchard, as the ‘Primary Value Problem’, his formulation of the problem doesn't mention value at all. Instead, he presents it as the question of why we tend to prefer knowledge over mere true belief. There is no indication that Bondy considers his formulation to be in any way unusual. Indeed, the reason I have taken two of my examples from well-known internet encyclopedias of philosophy is precisely to emphasise how standard it is to formulate Platonic value problems in these different ways – as questions about our desires, questions about our preferences, and as questions about the aims of our activities.Footnote ¹⁴

Why do philosophers formulate Platonic value problems in these different ways? The questions about our desires, preferences and aims clearly aren't equivalent to questions about the value of knowledge. That agents desire X and not Y doesn't entail that X is more valuable than Y, or even that agents value X more highly than Y. The same point applies to seeking X rather than Y – one might seek X rather than Y not because one takes X to be more valuable than Y, but because one takes X to be more easily attainable than Y. Likewise, an agent can prefer X to Y without valuing X more highly than Y. Some preferences are personal, and we take them to be so. I prefer mayonnaise to ketchup on chips, but I don't take mayonnaise to be ‘better’ than ketchup on chips in an objective sense. Since claims that we seek, desire or prefer X rather than Y so obviously do not entail that X is more valuable than Y, the reason philosophers formulate Platonic value problems in these different ways can't be that they mistakenly take questions about our desires, preferences and aims as enquirers to be equivalent to contrastive questions about the value of knowledge. Something else is going on.

A much more plausible proposal is that philosophers formulate Platonic value problems in these different ways because they make what I am calling ‘the value assumption’. They assume, that is to say, that contrastive questions about our desires, preferences and aims as enquirers are to be answered by appealing to claims about the superior value of the things desired, preferred or aimed at. So, for instance, the question of why we desire knowledge and not just true belief is to be answered by explaining why knowledge is more valuable than true belief. Once this assumption is in play, differences between contrastive questions about our desires, preferences and aims as enquirers and questions about the comparative value of the states desired, preferred, or aimed at are apt to seem unimportant. Furthermore, since this assumption is very natural, it isn't surprising to find that those who make it fail to explicitly articulate or defend it.

Given the significance of the value assumption for the argument of this paper, it is worth stating in a canonical form:

One issue we need to consider about the value assumption is whether it is true. I turn to this task in §3, where I outline what can be said in its favour. First, though, I want to say more about the significance of the value assumption – in particular, about the role of the assumption in motivating Platonic value problems.

2. The source of Platonic value problems

When philosophers discuss Platonic value problems, they typically don't say much about the motivation for accepting the claim that knowledge is more valuable than true belief-involving states necessary but insufficient for knowledge. Typically, this claim is an unquestioned presupposition of discussion; if it is commented on at all, it is simply asserted that we have an ‘intuition’ that it is true. Miranda Fricker introduces the Meno problem in this way:

If the argument of §1 is correct, this way of presenting Platonic value problems is misleading. Philosophers have not felt compelled to accept the claim that knowledge is more valuable than states implicating true belief but falling short of knowledge simply because they have an intuition that it is true. They have also felt compelled to accept this claim because they make the value assumption.

Once this point has been made, it is natural to ask a further question: is there any intuition that knowledge is more valuable than true belief independently of the value assumption? Or have philosophers only found this claim intuitive because of the influence of the value assumption? How we answer this question will fundamentally affect our conception of the status of Platonic value problems. If there is no intuition that knowledge is more valuable than true belief that is independent of the value assumption, then the fundamental problems in this area are not Platonic problems to do with the comparative value of states, but rather contrastive questions to do with the desires, preferences and aims of enquirers. Platonic value problems will only arise given the further assumption that the way to answer these contrastive questions is by appealing to claims about the comparative value of the state of knowledge.

My own view is that there is no intuition that knowledge is more valuable than true belief-involving states falling short of knowledge independently of the value assumption. I cannot prove that this claim is true. If someone insists that they have an ‘intuition’ that knowledge is more valuable than true belief that is wholly independent of the value assumption, I cannot refute them. However, I would argue that the onus is hardly on me to do so. No one should deny that philosophers have felt compelled to accept the claim that knowledge is more valuable than true belief partly because they have made the value assumption. The examples given in §1, together with the naturalness of the value assumption itself, show this to be so. Why, then, should we think that we also have a wholly independent intuition that knowledge is more valuable than true belief? If one took oneself to have such an intuition, how would one know that it was independent of the value assumption? Until answers to these questions are forthcoming we are justified in meeting blunt assertions of such intuitions with scepticism.Footnote ^16, Footnote ¹⁷

The question of whether the value assumption is true is therefore more important than it may initially have appeared. What is at stake is not merely how we are to approach contrastive questions about our desires, preferences and aims as enquirers – questions interesting enough in their own right – but the deeper belief that Platonic value problems are among the central questions of epistemology. Furthermore, once the role of the value assumption in motivating Platonic value problems has been exposed, the significance of observations about the respects in which knowledge appears to be no better than true belief (and other states falling short of knowledge) changes. Rather than deepening Platonic value problems, such observations might instead be taken as indications that the value assumption is false. Perhaps these problems only arise because natural assumptions about how we are to explain the desires, preferences and aims of agents break down in the case of knowledge and true belief. Further investigation of the value assumption is clearly required.

3. Seeking, Desiring, Preferring, and Valuing

A natural line of thought supports the idea that questions about our desires, aims and preferences as enquirers are to be settled by appealing to considerations about the comparative value of states. This line of thought begins with the observation that ordinary human beings engage in activity with the aim of finding out – i.e. acquiring knowledge of – the answers to questions. Such questions needn't be deep or profound, or imply ‘enquiry’ in a grand sense of the term; the sort of thing I have in mind is exemplified just as well by looking out of the window to see what the weather is like, or turning on the news to find out the headlines. Given that we engage in activity that aims at the acquisition of knowledge (or, as I will sometimes put it, that we search for knowledge), it seems undeniable that we sometimes want to know the answers to questions rather than merely wanting to arrive at true beliefs about them. It therefore seems legitimate to ask why we want to know the answers to questions rather than merely wanting to arrive at true beliefs about them. A natural response to this question is that we sometimes desire knowledge and not just true belief because, on those occasions, we prefer knowledge to true belief. Once it is admitted that we have this preference, it is hard to avoid the conclusion that we value knowledge more highly than true belief (at least on those occasions where we desire knowledge and not just true belief). Although, as I pointed out earlier, some preferences are personal and do not express valuations on the part of the subject, the suggestion that the preference for knowledge over true belief might be a personal preference in this sense is obviously implausible. This preference appears to be universal, which suggests that there are interpersonally available reasons for preferring knowledge to true belief, or at least that we take there to be such reasons. In other words, the universality of this preference implies that we value knowledge more highly than true belief. That we value knowledge more highly than true belief is, in turn, evidence that knowledge really is more valuable than true belief. Admittedly, this evidence isn't conclusive – it is possible that our valuation is mistaken. Nevertheless, in the absence of an explanation of how we could have been led astray in this matter, the presumption should be that we are not, and that knowledge genuinely is more valuable than true belief.

This line of thought can be represented by a simple diagram:

That is, starting from an observation about what we seek (e.g. that we seek knowledge and not just true belief), the line of thought infers a claim about what we desire (we desire knowledge and not just true belief), thence a claim about what we prefer (we prefer knowledge to true belief), thence a claim about what we value (we value knowledge more highly than true belief), before terminating in a claim about value (knowledge is more valuable than true belief). Parallel considerations can also be used to support the claim that knowledge is more valuable than states intermediate between true belief and knowledge – the presupposition of the secondary value problem.

Making this line of thought explicit crystallises the task of resisting the value assumption. If we wish to reject that assumption (and hence avoid the Platonic value problems that come in its wake), but accept the initial observation that we sometimes engage in activity with the aim of acquiring knowledge of the answers to questions,Footnote ¹⁸ then we have basically four options, corresponding to the four arrows in the diagram. The first is to argue that, although we sometimes search for knowledge, this isn't to be explained by appealing to the idea that we desire knowledge and not just true belief. The second is to argue that, although we desire knowledge and not just true belief, this isn't because we prefer knowledge to true belief. The third option is to argue that, although we prefer knowledge to true belief, this isn't because we value knowledge more highly than true belief. The fourth option is to argue that, although we value knowledge more highly than true belief, we are mistaken in doing so, and knowledge is really no more valuable than true belief.

The second option is the one I will pursue here. A number of different explanations that make no appeal to a preference on our part for knowledge over true belief might be given of why we desire knowledge and not just true belief. The rest of this essay will evaluate one such explanation, taken from Bernard Williams's discussion of Descartes's method of doubt. Although I will argue that this explanation fails – indeed, that it is prey to decisive objections – I will also suggest that seeing why it fails is instructive for future work in this area.

4. The means to true belief

Williams's explanation focuses on the methods a rational enquirer must use to arrive at the truth. According to Williams, reflection on these methods shows that a rational enquirer who wants the truth about the answer to some question must want nothing short of knowledge of the answer to that question. If this is correct, then there is no need to appeal to a preference for knowledge over true belief to explain why enquirers desire knowledge and not just true belief. That enquirers want to know is rather a consequence of rational constraints on the pursuit of truth in ‘a non-magical world’ (40/25).Footnote ¹⁹

The starting point for Williams's explanation is an enquirer, A, who wants the truth about certain questions. As Williams observes, such questions can take various forms – ‘who…?’, ‘what…?’, ‘where…?’, etc. – but, for the sake of simplicity, Williams focuses on the case in which A wants the truth on a question of the form whether p. The first issue Williams considers is what A wants in wanting the truth about whether p. He suggests that, at the very least, A wants to have a true belief about whether p. In other words, A wants, at the very least, to be in the following state of affairs:

1. (i) if p, A believes that p, and if not-p, A believes that not-p

Is a true belief about whether p all that A wants in wanting the truth about whether p? Williams thinks not. Since A wants (and not merely wishes) to be in the state of affairs specified in (i), it follows that A must adopt purposive means to bring about (i). These means, Williams says, must consist in A using a method of arriving at a belief about whether p such that A is unlikely to arrive at a belief about whether p by that method unless that belief is a true one. In this respect ‘collecting’ true beliefs contrasts with collecting prehistoric flints. One method for collecting (only) prehistoric flints would be to collect as many flints as possible, and then to investigate which, if any, were prehistoric, discarding those which weren't. An analogous procedure for collecting true beliefs ‘would be not just inefficient, but incomprehensible’ (39/25), because one cannot acquire the belief that p whilst remaining neutral about whether p is true.

Williams introduces ‘E’ as a label for the property a belief has when it was produced by a method unlikely to produce a false belief about whether p. If A succeeds in arriving at a true belief about whether p by a method of this kind, A will be in the following state of affairs:

1. (ii) if p, A believes that p, and if not-p, A believes that not-p, and in either case A’s belief has the property E

Furthermore, Williams says, (ii) not only specifies the state of affairs A will end up in if his enquiries are successful, but also specifies the content of A’s desire for the truth about whether p. Williams takes this further point to be a consequence of the fact that A knows that he can only arrive at a true belief about whether p by engaging in enquiry.Footnote ²⁰ If this is right, we can now say that, in wanting the truth about whether p, A wants at least to be in the state of affairs described in (ii).

Is being in (ii) sufficient for A to know whether p? If it is, Williams claims, we will have shown why it is that, in wanting the truth, A wants to know. Now, Williams takes the insufficiency of (ii) for knowledge to be a moral of Gettier cases. However, he goes on to suggest that, in broad terms, the ingredient required for knowledge missing in Gettier cases is clear enough: ‘the truth of the belief and the use of the method are detached from one another’ (44/29). As a rational enquirer, Williams thinks, A cannot want the truth of his beliefs to be ‘detached’ from the methods he uses to arrive at those beliefs. Thus, he claims, we can further specify the property of beliefs, E, as it figures in the description of the state of affairs A wants to end up in, as the property a belief has when it was ‘appropriately produced in a way such that beliefs produced in that way are generally true’, ‘where ‘appropriately’ means that the truth of the belief is not accidental relative to what it is about the way of its production which makes that a generally reliable way’ (45/31). With E so construed, being in the state of affairs described in (ii) ‘surely is sufficient’ for A to know whether p. So, Williams concludes, ‘starting merely from the idea of pursuing truth in a non-magical world, and so of the truth-seeker's using methods of enquiry, we do arrive at the conclusion that the search for truth is the search for knowledge’ (45/31).Footnote ²¹

In my view, Williams's argument is riddled with problems. Rather than attempting to catalogue all of them, I will focus on what I take to be the central difficulty with his approach. The essence of that approach is to motivate claims about the content of an enquirer's desire for truth by appealing to claims about rational constraints on the pursuit of truth.

An immediate problem with the way Williams develops this approach concerns his claim that an enquirer who wants to be in (i) ‘must adopt purposive means to get into (i)’ (39/24). There appears to be no sense of ‘must’ on which this is true. It is certainly possible to have a desire without acting in ways designed to secure its satisfaction. Moreover, the fact that one has a desire doesn't rationally oblige one to adopt purposive means to bring about what one desires – a person on a diet who wants an ice cream isn't rationally obliged to buy one. A response to this objection is to say that, rather than merely wanting to be in (i), the enquirer intends to get himself into (i), and one must pursue the means to one's ends. (Williams's mistake, according to this response, is to think that merely desiring something suffices for it to be an end.)Footnote ²² However, if one responds to the objection in this way, one can no longer hope to motivate claims about the content of an enquirer's desire for truth by appealing to claims about rational constraints on the pursuit of truth: it isn't the case that an enquirer who merely desires the truth must adopt purposive means to attain it.

Let's set this problem aside. Even if we cannot hope to show that the desire for truth is the desire for knowledge by appealing to rational constraints on the pursuit of truth, it might still be held that we can appeal to this type of consideration to show why an enquirer who intends to get the truth intends to arrive at a state of knowledge. If we can do so, then the line of thought that supports the value assumption (see §3) falls at the first hurdle: we do not need to appeal to the idea that enquirers desire knowledge and not just true belief to explain why they seek knowledge and not just true belief. That they seek knowledge is instead a consequence of rational constraints on the pursuit of truth.

It is here, however, that we arrive at the fatal flaw in Williams's argument. It is simply a mistake to infer from the fact that an agent has a certain aim, and intends to achieve that aim by certain means, that the agent's aim is, moreover, to achieve what they originally aimed at by those means. That this inference is valid is a basic assumption of Williams's approach, for it is only if we can (so to speak) build the means an enquirer must use to achieve his aim into the aim itself that considerations about those means can be used to amplify a minimal conception of the enquirer's aim.

A familiar example shows this inference to be invalid.Footnote ²³ Suppose I intend to drink a good espresso. Rather than going to a coffee shop, I decide to make one for myself. Recognising that the only consistent way to make a good espresso is to use a trustworthy coffee machine, I form the further intention of using a reliable coffee machine to make one. Although I aim to drink an espresso, and I intend to achieve that aim by making an espresso with a reliable coffee machine, it doesn't follow that I have the further aim of drinking an espresso made by a reliable coffee machine. To see this, consider how I would react if I were to make an espresso using an unreliable machine that happened, on this occasion, to produce a good one. I wouldn't react by thinking that I hadn't achieved my aim; on the contrary, I would think that I had (luckily) achieved my aim, despite failing to use my intended means.

This simple point about aims is fatal to Williams's argument. Even if we grant Williams the (large) claim that an enquirer who aims to arrive at a true belief about whether p must intend to use means of a kind that would suffice for a true belief they produced to constitute knowledge of whether p, it doesn't follow that the enquirer's aim is to arrive at knowledge of whether p. To draw this conclusion is to illegitimately build the means the enquirer must use to achieve his aim into the aim itself. Williams's argument therefore fails to show that ‘the search for truth is the search for knowledge’.

Here it may be objected that, even if Williams fails to show that the search for truth is the search for knowledge, his discussion establishes the still significant result that an enquirer who aims to get the truth on some question must intend to arrive at the truth using means that suffice for them to know the answer to that question.Footnote ²⁴ There are two points in response to this objection. The first is that although I was prepared to grant Williams the premise that an enquirer seeking a true belief must intend to arrive at a true belief via means that suffice for knowledge, this was not because I think that this premise is plausible. In fact, I think that even this weaker claim is dubious.Footnote ²⁵ Secondly – and more importantly – regardless of whether Williams succeeds in establishing the weaker claim, the present interest of his argument arose from the possibility of using it to undermine the line of thought supporting the value assumption (see §3). To undermine this line of thought, we needed to find a way of explaining why we search for knowledge and not just true belief, or why we desire knowledge and not just true belief, that didn't appeal, in the end, to the claim that knowledge is more valuable than true belief. The claim that an enquirer searching for true beliefs must intend to arrive at a true beliefs via means that suffice for knowledge doesn't undermine this line of thought. As I emphasised above, even if we grant this claim, it doesn't follow that an enquirer cannot search for true beliefs without searching for knowledge. Hence, regardless of whether the weaker claim would satisfy Williams, it is too weak to satisfy our present requirements.

If the general approach embodied in Williams's argument is fatally flawed, what follows for the task of resisting the value assumption? One thing that follows, I think, is that if we are to explain why enquirers desire knowledge and not just true belief in a way that doesn't appeal to a preference for knowledge over true belief, we mustn't do so by trying to explain why a desire for knowledge is consequent on some more minimal specification of the enquirer's desire for the truth. If one does approach the task in this way, it is hard to see what one can appeal to besides the means an enquirer would use to arrive at the truth. For the reasons just given, this appeal is bound to fail.

The point that we should reject the approach that tries to build up from some more minimal specification of the enquirer's desire for truth is reinforced by the observation that there is no good reason to specify the content of the enquirer's basic desire for truth as the desire for true belief and not as the desire for knowledge. That there is reason for doing so might be thought to be shown by the following reflection: ‘The enquirer wants to have the truth about the answer to some question – let us suppose, a question of the form whether p. If someone has a true belief about whether p, then he has the truth about whether p; but having a true belief about whether p is insufficient for knowing whether p. Hence, an enquirer who wants the truth about whether p at most wants a true belief about, rather than knowledge of, whether p.’ This reflection assumes that the enquirer starts out with a purely generic desire for the truth – i.e. a desire which isn't more specific in the manner of being (e.g.) a desire for knowledge or a desire for true belief – and that we can then identify the minimal state that would satisfy this generic desire by identifying the minimal state being in which suffices for one to in some sense ‘possess’ the truth. But this assumption is false. There is no such thing as a purely generic desire for the truth: if one desires the truth, one must always desire it in some more specific way – one must desire to know the truth, or to have a true belief, or to guess the truth, and so on. For this reason the question of ‘why one who wants the truth should want to know’ (37/23) raises no real issue. The correct response to this question is to ask the questioner to specify what one wants in wanting the truth. If what one wants is knowledge, then there is no further question of why, in wanting the truth, one wants to know. If, on the other hand, what one wants falls short of knowledge, then one doesn't want to know, so there can be no question of why one should want to know.

Although this point disposes of Williams's question about enquiry, it doesn't dispose of the question of why we desire knowledge and not just true belief: this question doesn't assume that our desire for true belief is more basic than our desire for knowledge. However, it does show that, in answering that question, there is no need to explain why an enquirer who is initially specified as wanting only a true belief should want to know. This is a good thing, for (as I argued above) it is hard to see how such an explanation could possibly succeed. In attempting to explain why we desire knowledge and not just true belief in a way that makes no appeal to a preference for knowledge over true belief, then, we shouldn't take as our starting point an enquirer who merely wants a true belief about whether p, and ask why such an enquirer should want to know. Instead, we should focus on the following question: why, typically, do we not merely desire to arrive at true beliefs about the answers to questions? What would be odd (perhaps even incoherent) about wanting to arrive at a true belief about whether p without this being because, more fundamentally, one wanted to know whether p? Pursuing this question is, I believe, the most promising way of resisting the line of thought that supports the value assumption, and therefore of avoiding the Platonic value problems that accompany it.Footnote ²⁶